Eos,Vol. 86, No. 1, 4 January 2005 With support from the U. S. National Science References Oganov,A. R., and S. Ono (2004),Theoretical and Foundation (NSF), a multidisciplinary work- experimental evidence for a post-perovskite phase Badro, J., J.-P.Ruef, G.Vankó,G. Monaco, G. Fiquet, and ″ shop organized by the Cooperative Institute of MgSiO3 in Earth’s D layer, Nature, 430, 445–448. F.Guyot (2004),Electronic transitions in perovskite: for Deep Earth Research (CIDER) was held at Shim, S.-H.,T.S. Duffy,R. Jeanloz, and G. Shen (2004), Possible nonconvecting layers in the lower man- Stability and crystal structure of MgSiO perovskite the Kavli Institute for Theoretical Physics of 3 tle, Science, 305, 383–386. to the core-mantle boundary, Geophys. Res. Lett., the University of California, Santa Barbara on Humayun, M., L. Qin, and M. D.Norman (2004), Geo- 31, L10603, doi:10.1029/2004GL019639. 12 July–6 August 2004 (http://online.kitp.ucsb. chemical evidence for excess iron in the mantle Sidorin, I., M. Gurnis, and D.V.Helmberger (1999), edu/online/earth04/). Part of the workshop beneath Hawaii, Science, 306, 91–94. Dynamics of a phase change at the base of the focused on the post-perovskite phase transi- Iitaka,T., K. Hirose, K. Kawamura, and M. Murakami mantle consistent with seismological observations, (2004),The elasticity of the MgSiO post-perovskite tion, and highlighted the emerging multidisci- 3 J. Geophys. Res., 104, 15,005–15,023. phase in the Earth’s lowermost mantle, Nature, plinary challenges. Tsuchiya,T., J.Tsuchiya, K. Umemoto, and R. M. 430, 442–445. CIDER provides a promising community Wentzcovitch (2004a), Elasticity of post-perovskite Lay,T., E. J. Garnero, and Q.Williams (2004), Partial MgSiO ,Geophys.Res.Lett.,31,L14603,doi:10.1029/ organization for communication and collabo- melting in a thermo-chemical boundary layer at 3 2004GL020278. ration across the relevant disciplines, and the the base of the mantle, Phys. Earth Planet. Inter., Tsuchiya,T., J.Tsuchiya, K. Umemoto, and R. M.Wentz- NSF Program for Cooperative Studies of the 146, 441–467. covitch (2004b), Phase transition in MgSiO per- Earth’s Deep Interior is a key source of research Mao,W.L., G. Shen,V.B. Prakapenka,Y.Meng, 3 ovskite in the Earth’s lower mantle, Earth Planet. Sci. A. J. Campbell, D.L. Heinz, J. Shu, R. J. Hemley,and support. Special sessions recently held at the Lett., 224, 241–248. 2004 AGU Fall Meeting, and upcoming at the H.-K. Mao (2004), Ferromagnesian postperovskite ″ 2005 Joint Assembly,are another valuable silicates in the D layer, Proc. Natl.Acad. Sci. U.S.A.,101, 15,867–15,869. means by which to inform and catalyze inter- Matyska, C., and D.A.Yuen (2004),The importance of Author Information actions across the community. radiative heat transfer for superplumes with a deep mantle phase transition,Earth Planet. Sci. Thorne Lay,University of California, Santa Cruz; Lett., 125, 255–266. Dion Heinz, University of Chicago, Ill.; , Murakami, M., K. Hirose, K. Kawamura, N. Sata, and Y. Institute of and Planetary Physics, Acknowledgments Ohishi (2004), Post-perovskite phase transition in Scripps Institution of Oceanography,University of

MgSiO3, Science, 304, 855–858. California, San Diego; Sang-Heon Shim, Massachu- The authors thank Barbara Romanowicz Nakagawa,T.,and P. J.Tackley (2004), Effects of a per- setts Institute of Technology,Cambridge; and Jun and Adam Dziewonski for organizing the 2004 ovskite-post perovskite phase change near core- Tsuchiya,Taku Tsuchiya, Renata Wentzcovitch, and CIDER workshop, which was partially support- mantle boundary in compressible mantle David Yuen,University of Minnesota,Minneapolis ed by NSF under grants PHY99-07949 to the convection, Geophys. Res. Lett., 31, L16611, For additional information, contact T.Lay; Kavli Institute and EAR-0215587 to CIDER. doi:10.1029/2004GL020648. E-mail: [email protected].

Although the notion is widespread among New Approaches for climate scientists that there were no operational upper air measurements before about 1948, this is not the case (Figure 1). Radiosonde Extending the Twentieth observations have been made since the mid- 1930s. Prior to the radiosonde era,weather bal- Century Climate Record loons with graphical registering devices were used.Even more common were kite soundings or aircraft measurements up to 4 or 5 km alti- PAGES 2,6 transition to the “modern era”of climatology, tude.Pilot balloons have been launched rou- Studying twentieth century climate is a key which started after World War II.The reasons tinely since the early twentieth century to to understanding future climate change. Rela- are manifold and include military secrecy; obtain information on upper level winds.In tively little is still known,however,about climate interrupted international collaboration; politi- addition to meteorological measurements, variability in the first half of the century.Much cal and institutional changes during and fol- spectrographical total ozone observations could be learned from the relatively large cli- lowing the war; and,sometimes,simply neglect. reaching back to the 1920s can be used to matic variations that occurred during that first Yet,these data can still be found today on derive indirect information on stratospheric half,including the decade-long “Dust Bowl” paper in various meteorological archives.With dynamics. droughts of the 1930s and the warming of the new numerical and statistical techniques The total amount of data is small by current Arctic from 1920 to 1945. becoming available,these archives now could standards,but it is non-negligible.It is estimated Poor digital data availability prior to around be fruitfully mined for climate research. that several million pilot balloons were launched 1948 has hindered previous work to understand Data availability is comparably good for prior to 1948, and there were several hundred these important climatic variations. meteorological observations at the Earth’s sur- thousand radiosonde ascents and aircraft Several projects now are focusing on digitizing face, which have been used continuously to flights (Figure 1). earlier manuscript observations to create three- study past climate variability.Several ongoing Significant fractions of these data are dimensional,gridded meteorological data sets projects are increasing the data quality and currently being digitized and processed by for the first half of the twentieth century. quantity [e.g.,Worley et al.,2005].Surface data, organizations such as the U.S.National Climate These data sets are likely to provide further however,do not suffice to fully understand Data Center (NCDC), the National Center for insights into processes governing interannual- the mechanisms governing large-scale climate Atmospheric Research (NCAR),and the World to-interdecadal large-scale climate variability. variability. Data Center (WDC) for Meteorology in Meteorologists, geophysicists, navy pilots, Upper air data are needed for accurate Obninsk, Russia.Additional work is done by ship crews,and numerous volunteer observers descriptions of important dynamical features scientists at the Swiss Federal Institute of collected enormous amounts of atmospheric such as the positions of the jet streams, the Technology (ETH) Zürich in the framework of data in the first half of the twentieth century, planetary wave structure, and the strength of a Swiss National Science Foundation project sometimes under extreme and dangerous the stratospheric polar vortex.Yet,gridded led by the first author. conditions.A large fraction of these data, upper air data sets currently are available only Re-evaluating historical upper air data is especially upper air data, never made the for the second half of the twentieth century; demanding work.After digitizing the numbers they are based largely on radiosonde data, from paper,the data need to be corrected for BY S. BRÖNNIMANN,G.P.COMPO,P.D.SARDESHMUKH, and since 1973, on radiosonde and satellite various instrumental errors,a task made difficult R. JENNE, AND A. STERIN data (Figure 1). by the lack of good background information. Eos,Vol. 86, No. 1, 4 January 2005

Environmental Sciences, sponsored jointly by NOAA and the University of Colorado,Boulder]/ CDC) in Boulder,Colorado, new techniques such as the “ensemble square root-filter” (EnSRF) [Whitaker and Hamill, 2002] have been developed and tested. Figure 2 shows Northern Hemispheric 500-hPa (hectopascals) geopotential height fields for 14 December 2001,0000 UTC,obtained from a conventional assimilation of all available data (surface, upper air,and satellite; left panel), and an EnSRF assimilation using only a limited number of surface pressure observations (right panel), mimicking the network of land stations and marine observations of the year 1915 (black dots) [Whitaker et al., 2004]. The two fields are very similar; their differ- ence (the error of the “reanalysis”) is of com- parable magnitude to current 2.5-day forecast errors. Overall, the results of Whitaker et al. suggest that useful upper air circulation analyses up to the middle troposphere may already be feasible with the available digitized data, even for times prior to any upper air observations. Better results could be expected if historical upper air data were also to be included.The goal of the efforts at NOAA-CIRES/CDC is to produce a new reanalysis data set for the entire twentieth century to present. Another approach to deriving fields from scattered upper air observations is reconstruc- tion,i.e., by using statistical relationships derived from the modern era.The technique has been widely used in climate research to reconstruct surface fields on a monthly to annual scale [Luterbacher et al., 2004]. Statistical reconstruction is much simpler and less costly than data assimilation, but does not have the ability to represent high-res- olution features. Nevertheless, relatively good quality monthly fields of temperature and geopotential height up to 100 hPa have recently Fig. 1. (top) Upper air data prior to 1948 that is currently available in digital format from the data been reconstructed for the extratropical sets TD52 and TD53 (NCAR, http://dss.ucar.edu/docs/papers-scanned/papers.html,documents Northern Hemisphere back to 1939 [Brönni- RJ0167, RJ0168), UA39_44 [Brönnimann, 2003], and CARDS (Comprehensive Aerological Refer- mann and Luterbacher, 2004]. ence Data Set) (http://www.ncdc.noaa.gov/oa/climate/cards/cards_homepage.html). (bottom As an example, 300-hPa geopotential height left) A schematic of the types of upper air and surface observations performed during the twenti- anomalies for March 1941 are shown in Figure eth century.The top part shows the currently available and planned gridded upper level data 3 together with the upper air stations used for sets. (bottom right) Launch of radiosonde at Washington Airport,Washington,D.C., 7 May 1936 (source: National Oceanic and Atmospheric Administration Photo Library,U.S. Department of the reconstruction. In view of these results, the Commerce). data now being digitized are expected to yield reliable reconstructions up to the upper tro- posphere, and regionally even up to the lower In fact,even for the second half of the century, from a numerical weather prediction model stratosphere, back to the mid-1920s. where such information is available,the quality to yield a physically consistent,“optimal”rep- It is hoped that the new upper air data sets of radiosonde data remains a major challenge, resentation of the data every few hours. Such and improved surface data,which will become for example,when attempting to correct for reanalysis data sets are the standard in atmos- available within the next few years,will lead to changes in instruments and practices [e.g., pheric research today,the most important a better understanding of important climate Lanzante et al., 2003]. examples being NCEP-NCAR Reanalysis and variations such as the Dust Bowl [Schubert et al., Despite these problems, preliminary results the European Centre for Medium-Range 2004] of the 1930s,the global climate anomaly [Brönnimann, 2003] suggest that useful upper- Weather Forecast Reanalysis Projects ERA15 of the early 1940s,and the Arctic warming air information for the extratropical Northern and ERA40 (see Figure 1). from 1920 to 1945 [Polyakof et al.,2003].In Hemisphere could be obtained back to around Because much less data are available in the 1925. pre-reanalysis period (prior to 1948),an assim- conjunction with the already available reanalysis ilation of historical data is not expected to data sets for the second half of the twentieth century,they will offer a new look at climate From Point-Data to Three-Dimensional Grids provide the same quality product as current data sets. However,more powerful assimilation variability during the entire century. Several approaches can be used to derive schemes optimized for sparser observations The authors appreciate contributions by gridded spatial fields from this information. can be used. members of the AGU community.Eos readers Data assimilation combines a statistical filtering At the Climate Diagnostics Center (NOAA- having information on pre-1948 upper air technique together with background information CIRES [Cooperative Institute for Research in observations,stations,instruments,calibration, Eos,Vol. 86, No. 1, 4 January 2005

Fig. 3.The 300-hPa geopotential height anom- Fig.2.A 500-hPa geopotential height analysis for 14 December 2001,0000 UTC (contour interval alies (in geopotential meters, with respect to are 50 m).The NCEP-NCAR Reanalysis using all available observations is shown on the left.The 1961–1990) for March 1941, obtained from EnSRF analysis using the simulated 1915 surface pressure observations network (black dots) is statistical reconstructions [Brönnimann and shown on the right (reprinted from Whitaker et al. [2004], Copyright 2004 American Meteorolog- Luterbacher, 2004].Black dots mark the upper ical Society).The root mean square difference between the left and right panels is 33 m, and the air stations used in the reconstructions (in anomaly pattern correlation is 0.96. addition to upper-air data, 100 surface temper- ature series as well as sea level pressure fields were used). Lighter shaded areas denote a technical details, measurement and reporting Luterbacher,J., D.Dietrich, E. Xoplaki, M. Grosjean, and H.Wanner (2004),European seasonal and low reconstruction skill (RE < 0.2) [see Brön- procedures, etc., are invited to contact the first nimann and Luterbacher, 2004]. author: [email protected]. annual temperature variability,trends and extremes since 1500, Science, 303, 1499–1503. Polyakof, I.V.,et al. (2003),Variability and trends of air temperature and pressure in the maritime Arctic, Worley,S. J., S. D.Woodruff, R.W.Reynolds, S. J. Lubker, References 1875-2000, J. Clim., 16, 2067–2077. and N. Lott (2005), ICOADS Release 2.1 data and Schubert, S. D., M. J. Suarez, P. J. Pegion,R. D.Koster, products. Int. J. Climatol., submitted. Brönnimann, S. (2003),A historical upper-air data set and J.T.Bacmeister (2004), On the causes of the for the 1939-1944 period, Int. J. Climatol., 23, 1930s Dust Bowl, Science, 303, 1855–1859. 769–791. Whitaker,J. S., and T.Hamill (2002), Ensemble data Author Information Bronnimann, S., and J. Luterbacher (2004), Recon- assimilation without perturbed observations,Mon. structing Northern Hemisphere upper level fields Weather Rev., 130, 1913–1924. S.Brönnimann,ETH Zurich,Switzerland; G.P.Compo during World War II, Clim. Dyn., 22, 499–510. Whitaker,J. S., G. P. Compo, X.Wei,and T.M. Hamill and P. D.Sardeshmukh, NOAA-CIRES CDC, Boulder, Lanzante, J. R., S.A. Klein, and D.J. Seidel (2003),Tem- (2004), Reanalysis without radiosondes using Colo.; R. Jenne, NCAR/UCAR, Boulder,Colo.; and A. poral homogenization of monthly radiosonde tem- ensemble data assimilation,Mon.Weather Rev., Sterin, Research Institute of Hydrometeorological perature data: Methodology, J. Clim., 16, 2224–2240. 132, 1190–1200. Information-World Data Centre,Obninsk, Russia

direction of ~140°E. However,they did these Did the 26 December 2004 Sumatra, Indonesia, so very slightly that the resultant signals have Disrupt the Earth’s Rotation as the so far eluded detection, even with today’s space geodetic technique capabilities.Worse Mass Media Have Said? still, these signals were buried in other signals that are orders of magnitude larger resulting from various other geophysical and climatic PAGES 1–2 the low-degree gravitational field.The algorithm causes occurring all the time. uses the normal-mode summation scheme by The answer to this question is a definite yes. This was the case at least up until recently. inputting the Harvard centroid-moment tensor But then again, the same is true of any earth- Previously,however,there had been two gigantic solution (courtesy of http://www.seismology. quake, large or small, or for that matter of any in the 1960s that had also been harvard.edu/CMTsearch.html),which represents worldly event that involves mass transport, geophysically modeled, namely,the 1960 the magnitude and focal mechanism of a given from atmospheric and ocean seasonality,to Chilean event and the 1964 Alaskan event. earthquake.The results are reported and melting of glaciers and tropical storms, to a They should have caused geodynamic changes bus driving around town.All one needs to updated on the Web site of the Special Bureau for Mantle of the International Earth Rotation that were large enough to be detected under convince oneself of this is to invoke the con- today’s observational capability,which was of servation of angular momentum and apply it and Reference Systems Service’s (IERS) Global Geophysical Fluids Center (http://bowie.gsfc. course lacking at the time. For example, the to the Earth system. Chilean earthquake should have shifted the The real question should be, Did this partic- nasa.gov/ggfc/mantle.htm/). Currently includ- North Pole toward ~115°E by about 23 mas, ular earthquake disrupt the Earth’s rotation to ed are 21,600 major earthquakes worldwide corresponding to ~70 cm, compared with a level large enough to be noticeable, or,tech- with magnitude greater than 5 since 1977. nically,observable? The answer is a sobering Their cumulative, coseismic geodynamic today’s subcentimeter measurement precision. hardly,but at the same time very exciting in effects show intriguing long-term trends for The corresponding change in LOD,on the other scientific implications. geophysicists to ponder.For instance, the hand, was only about -8 microseconds (µs),a Following Chao and Gross [1987; see also earthquakes collectively have an extremely few times below today’s detection level.The Chao and Gross,2000],we have been routinely strong tendency to make the whole Earth Alaskan earthquake should have changed the -11 calculating earthquakes’ coseismic effects in rounder and more compact in all directions, Earth’s oblateness J2 by +5.3 × 10 ,which would changing the Earth’s rotation (in both length shortening LOD.They have also been nudging take the postglacial rebound 2 years to “iron of day (LOD) and polar motion) as well as the mean North Pole position toward the out,”compared with today’s detectability level